Longitudinal Patterns of Cigarette Use Among South African High School Students

1
Longitudinal Patterns of Cigarette Use Among
South African High School Students Lori-Ann
Palen1, Edward A. Smith1, Linda L. Caldwell1,
Bethany C. Bray1, Alan J. Flisher2 1The
Pennsylvania State University, 2University of
Cape Town Funded by NIH R01 DA01749, NIH T32
DA017629-01A1
RESULTS
METHOD

• The optimal model (G2 118.74, 96 df) had four
  longitudinal smoking patterns, which were the
  same across both gender and treatment group
  (Figure 1).

• PARTICIPANTS
• 2,425 students from Mitchells Plain, South
  Africa
• Participating in a research trial of HealthWise
  South Africa (41 program, 59 control)
• Baseline demographics mean age 14 years 51
  female 86 mixed race

• MEASURES
• Seven semi-annual assessments, beginning in
  Grade 8
• At each assessment, dichotomous indicator of
  whether participant had smoked in his/her
  lifetime
• If yes, dichotomous indicator of past-month
  smoking
• If yes, quantity smoked in the past month
• Participant classified as regular smoker if
  he/she smoked 10 cigarettes in the past month

• Compared to the control group
• HealthWise boys were especially likely to be
  quitters.
• HealthWise girls were especially likely to be
  non-smokers (Figure 2).

Figure 2 Smoking group membership by gender and
treatment status

• ANALYSIS
• Repeated-measures latent class analysis (SAS
  PROC LCA) of regular smoking at each of seven
  time points
• LCA (Lanza, Flaherty, Collins, 2002) uses
  manifest variables to classify individuals
  according to underlying, unobservable constructs.
  
• Measurement parameters show the probability of a
  given response to an item (e.g., smoking at
  baseline), conditional on latent class.
• Group membership parameters show the probability
  of participants being classified in each latent
  class (i.e., longitudinal smoking pattern)
• Selection of a model with an optimal number of
  classes based on fit and parsimony
• Multiple groups analysis to determine whether
  measurement and group membership varied across
  gender and treatment group

DISCUSSION

• Initiation of regular cigarette use in this age
  group tended to occur by 9th grade therefore,
  prevention efforts should begin prior to this
  time.
• In the absence of intervention, smoking cessation
  is rare (4 of control group).
• There was evidence for the effectiveness of
  HealthWise in preventing smoking onset among
  girls.
• There was evidence for the effectiveness of
  HealthWise in promoting smoking cessation among
  boys. This suggests that there may be value in
  presenting prevention messages even after smoking
  onset.